Gas turbine transient model automatic correction method based on digital object fusion

Abstract

The invention relates to the technical field of gas turbine performance monitoring, and provides a gas turbine transient model automatic correction method based on digital object fusion, which comprises the following steps: step 1, carrying out data preprocessing on operation data of a gas turbine; 2, establishing a transient model of the gas turbine; step 3, determining key influence parameters having significant sensitivity to thermodynamic efficiency and hydrodynamic characteristics, and determining correction factors according to the key influence parameters; step 4, calculating the correction factor of the transient model of the gas turbine determined in the step 3; and 5, adjusting the gas turbine transient model by using the correction factor obtained by calculation to obtain a corrected gas turbine transient model. According to the invention, the accuracy and response reliability of dynamic adjustment of the model can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of gas turbine performance monitoring, and in particular to a method for automatically correcting a gas turbine transient model based on mathematical-physical fusion. Background Art

[0002] During operation, gas turbine (GT) components can degrade due to wear, fatigue, corrosion, or fouling. This can manifest itself as reduced efficiency and flow capacity, leading to a deviation between the actual operating state of the GT and the design state of the GT model. To address this issue, experts, scholars, and engineers typically incorporate component health indicators into performance models to represent the degree of equipment degradation. These health indicators are degradation factors for mass flow and efficiency, also known as health parameters in gas path analysis (GPA). The value of the health indicator ranges from 0 to 1, indicating the degree of performance degradation relative to optimal performance. A common method for monitoring GT performance degradation is to establish a real-time GT performance simulation model and simulate GT performance degradation using a characteristic map attenuation factor. By adjusting the characteristic map attenuation factor, the characteristics of the GT components can be adjusted so that the GT model output parameters are close to the actual data. However, due to the complex structure of gas turbines, the various components are interconnected and mutually restricted when working, and have strong coupling. Solving the attenuation factor with only one algorithm is not convincing enough. Therefore, it is very necessary to combine actual operating data and use modern optimization algorithms to automatically find the best solution to correct the model.

[0003] Traditional model correction methods mainly adjust the key parameters of the gas turbine mechanism model to make its output match the actual measurement data. They mainly include two types of technologies: the attenuation factor method based on component characteristic diagrams and the single optimization algorithm.

[0004] When correcting the component characteristic diagram attenuation factor method, the deviation between the actual performance of each component (such as the compressor and turbine) and the mechanism model is first identified through experimental data, and the attenuation factor is introduced to scale the key parameters of the characteristic diagram (such as efficiency and flow).

[0005] The single optimization algorithm correction method uses a numerical optimization algorithm to globally adjust model parameters. Common algorithms include least squares, gradient descent, and genetic algorithms. By constructing an objective function (such as the mean squared error or relative error between the model output and the measured data), the algorithm automatically searches for the optimal parameter combination.

[0006] The component characteristic diagram attenuation factor method is suitable for scenarios with clear subsystem boundaries and component-level test data. However, this method requires comprehensive component-level experimental data and suffers from high coupling effects between component factors, which can easily lead to repeated iterative parameter adjustments. Single optimization algorithm correction methods are prone to falling into local optimality for high-dimensional parameter systems, with computational efficiency decreasing exponentially with parameter dimensionality, and the correction results lack physical interpretability. A common limitation of both methods is their reliance on large amounts of measured data, which limits their ability to correct strongly nonlinear systems. Summary of the Invention

[0007] The present invention mainly solves the technical problems that the manual parameter adjustment method of the existing technology is inefficient and cannot adapt to dynamic changes in real time. It proposes an automatic correction method for the transient model of a gas turbine based on digital-physical fusion to achieve the purpose of improving the accuracy of the model's dynamic adjustment and the response reliability.

[0008] The present invention provides a method for automatically correcting a transient model of a gas turbine based on mathematical-physical fusion, which includes the following steps:

[0009] Step 1: preprocessing the gas turbine operating data;

[0010] Step 2, establishing a gas turbine transient model;

[0011] The gas turbine transient model includes: a compressor module, a combustion chamber module, a turbine module and a rotor dynamics module;

[0012] The performance of the compressor module is represented by a compressor characteristic curve diagram; the compressor characteristic curve diagram provides the compressor equivalent flow About compressor pressure ratio π c =p2 / p1, compressor reduced speed The functional relationship between the inlet guide vane angle f1 and the compressor efficiency η c About pressure ratio π c =p2 / p1, converted speed And the functional relationship f2 of the inlet guide vane angle IGV:

[0013]

[0014] The compressor reduced flow rate and compressor efficiency can be obtained through the functional relationship between the compressor pressure ratio, the compressor reduced speed and the IGV. By determining only three of the five parameters, the working state of the compressor can be determined.

[0015] The outlet temperature of the compressor module is:

[0016]

[0017] Where T1 and T2 are the compressor inlet temperature and compressor outlet temperature respectively; π c and η c are compressor pressure ratio and compressor efficiency respectively; k a is the air specific heat ratio;

[0018] After determining the compressor outlet temperature, the power consumed by the compressor module is:

[0019] P c =G c c pa (T2-T1)

[0020] Where G c is the compressor inlet air flow, c pa is the specific heat capacity of air at constant pressure;

[0021] The combustion chamber module characterizes the dynamic response characteristics of the outlet temperature and the variation law of the compression loss through the principles of conservation of mass and energy. In the gas turbine simulation, the mathematical model of the volume effect can be obtained by the mass conservation equation to obtain the following first-order differential equation:

[0022]

[0023] Where V is the size of the volume module, R is the gas constant, m is the mass of the gas in the volume module, and p is the volume of the gas. out and T out are the gas pressure and temperature at the outlet of the volume module, G in and G out The gas flow rates at the inlet and outlet of the volume module respectively;

[0024] The differential equations of the combustion chamber module dynamic process are expressed as follows:

[0025]

[0026]

[0027] In the formula, G2, G f G3 and G4 are respectively the air flow rate at the combustion chamber inlet, the fuel flow rate, and the gas flow rate at the combustion chamber outlet; p3 and T3 are respectively the pressure and temperature at the combustion chamber outlet; V is the combustion chamber volume; R g is the fuel gas constant, η B is the combustion efficiency of the combustion chamber; h2, H u and h3 are the enthalpy of air at the combustion chamber inlet, the lower calorific value of fuel, and the enthalpy of gas at the combustion chamber outlet, respectively;

[0028] Introducing the air load parameter as a correction factor, the following combustion chamber combustion efficiency formula is obtained:

[0029] ηB =-5.47×10 -11 L 5 +3.98×10 -8 L 4 -8.74×10 -6 L 3 +3.00×10 -4 L 2 -4.57×10 -3 L+99.7

[0030] Where: L is the air load parameter of the combustion chamber, which is defined as follows:

[0031]

[0032] The turbine module is based on the turbine characteristic curve, through the turbine expansion ratio π T and turbine efficiency η T Correlation calculation of turbine output power P T , and integrates the effect of cooling air flow on work, reflecting the performance under different working conditions; the turbine characteristic curve provides the turbine equivalent flow About turbine expansion ratio π T =p3 / p4, turbine equivalent speed The functional relationship f3, and the turbine efficiency η T About turbine expansion ratio π T =p3 / p4, turbine equivalent speed Function relationship f4:

[0033]

[0034] The outlet temperature of the turbine module is:

[0035]

[0036] Where T3 and T4 are the turbine inlet temperature and turbine outlet temperature respectively; π T and η T are turbine expansion ratio and turbine efficiency respectively; k g is the gas specific heat ratio; after determining the turbine outlet temperature, the turbine expansion work power is:

[0037] P T =G T c pg (T3-T4)

[0038] Where G T is the gas flow at turbine inlet, c pg is the specific heat capacity of the gas at constant pressure;

[0039] The rotor dynamics module is used to simulate and analyze the dynamic behavior of the gas turbine rotor. The rotor dynamics equation for the gas turbine rotor moment of inertia is:

[0040]

[0041] Since ω=πn / 30 and P=Mω, the rotor dynamics equation can be transformed into:

[0042]

[0043] Where ω is the rotor angular velocity, n is the gas turbine speed, I is the rotor moment of inertia, and P is T 、P c 、P G 、P f They are respectively the turbine work power, compressor work power, generator power and power generated by friction torque;

[0044] Step 3: determine the key influencing parameters that are significantly sensitive to the thermodynamic efficiency and fluid dynamic characteristics, and determine the correction factors based on the key influencing parameters;

[0045] Step 4, calculating the correction factor of the gas turbine transient model determined in step 3;

[0046] Step 5: Use the calculated correction factor to adjust the gas turbine transient model to obtain a corrected gas turbine transient model.

[0047] Furthermore, the operating data of the gas turbine includes thermal parameters, mechanical parameters, performance index data, control system data, and environmental and weather data;

[0048] The thermal parameters include compressor inlet temperature T1, compressor outlet temperature T2, compressor inlet pressure p1, compressor outlet pressure p2, and turbine outlet temperature T4;

[0049] The mechanical parameters include the rotational speed n of the rotor;

[0050] The performance index data includes the power generation P G , fuel flow G f ;

[0051] The control system data includes valve opening, IGV angle, and fuel distribution ratio f;

[0052] The environment and weather data include ambient temperature T0, ambient pressure P0, and ambient relative humidity HR.

[0053] Furthermore, the data preprocessing includes data cleaning;

[0054] The data cleaning includes the following processes:

[0055] Use sliding window interpolation or dynamic time warping to unify timestamps and synchronize data;

[0056] Perform data outlier detection and missing value interpolation;

[0057] The Bayesian wavelet packet method is used to reduce the noise of the data.

[0058] Furthermore, step 3 includes the following steps 301 to 303:

[0059] Step 301: determining key influencing parameters that are significantly sensitive to thermodynamic efficiency and fluid dynamic characteristics; the key influencing parameters include: compressor efficiency, combustor combustion efficiency, turbine efficiency, compressor inlet air flow, combustor total pressure recovery coefficient, and mechanical efficiency;

[0060] Step 302 , applying a dimensionless product term to the determined key influencing parameters, assigning a proportionality coefficient to the key influencing parameters, and obtaining a potential deviation range of the key influencing parameters under actual working conditions;

[0061] Step 303 : sort the key influencing parameters, and use the key influencing parameters whose potential deviation ranges meet the deviation threshold as correction factors for the gas turbine transient model.

[0062] Furthermore, step 4 includes the following steps 401 to 402:

[0063] Step 401: Establish the following calculation formula for the correction factor of the gas turbine transient model:

[0064] findx i i=1,2,3,...,m

[0065]

[0066] st 0.5≤x i ≤1.1

[0067] Among them, x i is the gas turbine transient model correction factor to be calculated, is the simulation result calculated by the transient mechanism model, is the actual operating data measured by the gas turbine, i and j represent the input parameters of the optimization problem and the number of corresponding variables, respectively, find means finding the best correction factor, min means minimization, obj means objective function, and st means constraint condition;

[0068] Step 402 : Calculate the correction factor of the gas turbine transient model using a gradient descent algorithm, a genetic algorithm, an ant colony algorithm, or a cuckoo search algorithm.

[0069] Furthermore, step 402 includes the following steps 4021 to 4024:

[0070] Step 4021, initializing key algorithm parameters in the gradient descent algorithm, genetic algorithm, ant colony algorithm, or cuckoo search algorithm;

[0071] Step 4022, performing fitness evaluation on the value of the objective function obj;

[0072] Gradient descent algorithm and cuckoo search algorithm directly use the objective function value as the fitness function;

[0073] Genetic Algorithm Use As a fitness function;

[0074] Ant colony algorithm selection As the fitness function, to select the path, where τ ij is the pheromone concentration of path i→j, η ij is the heuristic factor, α and β are the adjustment weight parameters;

[0075] Step 4023, updating parameters in the gradient descent algorithm, genetic algorithm, ant colony algorithm, or cuckoo search algorithm;

[0076] Step 4024: terminate when the termination condition is met, and obtain the value of the correction factor of the gas turbine transient model.

[0077] The present invention provides an automatic correction method for a gas turbine transient model based on the fusion of mathematics and physics. It constructs a hybrid correction framework based on the collaboration of mechanism and data. First, a component-level gas turbine transient model is established through a component decomposition method to improve the interpretability of the correction method. Then, sensitivity analysis is combined to screen key influencing parameters, determine the correction factor, and reduce the optimization dimension. Then, a variety of local optimization methods (gradient descent algorithm) and global optimization methods (genetic algorithm, ant colony algorithm, cuckoo algorithm) are used for optimization and adjustment. Through a dynamic method switching mechanism, an optimization method that meets actual conditions is selected to achieve a comprehensive improvement in model accuracy, computational efficiency, and interpretability.

[0078] By modeling the gas turbine according to the component-level decomposition method, the global correction problem is transformed into a multi-level subsystem optimization task, reducing the data dependency and coupling of the decoupling factor; the transient model accuracy of the gas turbine is improved by adopting methods such as predictive fitting of the characteristic curve of the compressor at low speed, correction of combustion efficiency of the combustor under variable operating conditions, and calculation of thermophysical properties. A parameter grouping strategy based on sensitivity analysis is adopted to prioritize the optimization of key factors and reduce the invalid search dimension; a variety of local and global optimization methods are introduced to replace traditional single algorithms. For example, in a small data scenario, the goal is to prioritize fast response, so a local optimization method with low computational complexity can be selected; in a large data scenario, the goal is to prioritize accuracy, so a global optimization method can be selected to improve the accuracy of modeling under all operating conditions. The model accuracy, real-time performance and robustness of the present invention are simultaneously improved, supporting the efficient design and control of complex systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 This is a flow chart for implementing the automatic correction method for the gas turbine transient model based on the fusion of mathematics and physics provided by the present invention.

[0080] Figure 2 is the low speed characteristic fitting curve of the compressor;

[0081] Figure 3 It is a flowchart comparing the core steps of four optimization algorithms in the context of model correction;

[0082] Figure 4 are the transient process inputs: a) ambient pressure b) ambient temperature c) fuel flow d) IGV angle;

[0083] Figure 5 It is the trend of the output power and speed of the gas turbine from cold start to steady state operation;

[0084] Figure 6 It is the influence of key parameters on gas turbine power output (Power) and turbine outlet temperature (TOT);

[0085] Figure 7 It is the error trend diagram of the genetic algorithm optimization algorithm in the basic load stabilization process as the number of iterations increases;

[0086] Figure 8 It is the comparison between the simulation result output and the actual data after transient process correction: a) compressor outlet pressure change b) turbine outlet pressure change c) compressor outlet temperature change d) turbine outlet temperature change e) output power change f) relative error between the simulation result and the actual data of each variable. DETAILED DESCRIPTION

[0087] To make the technical problems solved, the technical solutions adopted, and the technical effects achieved by the present invention more clearly apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, rather than all of the contents.

[0088] like Figure 1 As shown, an embodiment of the present invention provides a method for automatically correcting a gas turbine transient model based on digital-physical fusion, comprising the following steps:

[0089] Step 1: preprocess the gas turbine operating data.

[0090] The operating data of the gas turbine includes thermal parameters, mechanical parameters, performance index data, control system data, and environmental and weather data.

[0091] The thermal parameters include compressor inlet temperature T1, compressor outlet temperature T2, compressor inlet pressure p1, compressor outlet pressure p2, and turbine outlet temperature T4.

[0092] The mechanical parameters include the rotational speed n of the rotor.

[0093] The performance index data includes the power generation P G , fuel flow G f .

[0094] The control system data includes adjustment signals such as valve opening, IGV (inlet guide vane) angle, fuel distribution ratio f, etc.

[0095] The environment and weather data include ambient temperature T0, ambient pressure P0, and ambient relative humidity HR.

[0096] At present, data fusion of gas turbine power plants has shifted from single equipment monitoring to "unit-environment-grid" collaborative optimization.

[0097] The data preprocessing includes a series of processes such as data collection, data fusion, and data cleaning. Among them, the data cleaning includes the following processes:

[0098] 1) Use sliding window interpolation or dynamic time warping (DTW) to unify timestamps and perform data synchronization and alignment;

[0099] To address the asynchronous sampling problem of multi-source sensors (temperature, pressure, flow, etc.), sliding window interpolation or dynamic time warping (DTW) is used to unify timestamps and perform data synchronization alignment;

[0100] 2) Detect data outliers and interpolate missing values for the detected outlier data;

[0101] 3) The Bayesian wavelet packet method is used to reduce the noise of the data to improve the data quality.

[0102] Step 2: Establish a gas turbine transient model, which includes a compressor module, a combustion chamber module, a turbine module, and a rotor dynamics module.

[0103] The study of variable operating conditions of the compressor is the key to ensure the accuracy of the gas turbine model. The performance of the compressor module is usually represented by a compressor characteristic curve, which shows the compressor pressure ratio π at different speeds. c (ratio of outlet pressure to inlet pressure), compressor efficiency η c , rotor speed n (since the compressor and turbine of a single-shaft gas turbine are coaxially connected, compressor speed = turbine speed = rotor speed), compressor inlet air flow G c The relationship between the parameters. The compressor characteristic curve provides the compressor equivalent flow About compressor pressure ratio π c =p2 / p1, compressor reduced speed The functional relationship between the inlet guide vane angle f1 and the compressor efficiency η c About pressure ratio π c =p2 / p1, converted speed And the functional relationship f2 of the inlet guide vane angle IGV:

[0104]

[0105] The compressor reduced flow rate and compressor efficiency can be obtained through the functional relationship of the compressor pressure ratio, the compressor reduced speed and IGV. Only three of the five parameters need to be determined, and the working state of the compressor can be completely determined. Gas turbine manufacturers usually only provide high-speed steady-state characteristic data (such as design point, rated load performance), but in actual operation, the performance evolution in transient dynamic processes (such as startup, acceleration and deceleration, load mutation, etc.) involves low-speed conditions. The accuracy of low-speed characteristic prediction directly determines the reliability of the transient model, especially in scenarios involving rapid speed changes. However, low-speed experiments are risky (easy to enter the surge zone) and the data acquisition cost is high, resulting in a scarcity of low-speed experimental data. Therefore, the present invention uses high-speed data to migrate to low-speed conditions through domain adaptation, and the obtained low-speed characteristic fitting curve is as follows: Figure 2 shown.

[0106] The outlet temperature of the compressor module is:

[0107]

[0108] Where: T1 and T2 are the compressor inlet temperature and compressor outlet temperature respectively; π c and η c are compressor pressure ratio and compressor efficiency respectively; k a is the air specific heat ratio.

[0109] After determining the compressor outlet temperature, the power consumed by the compressor module is:

[0110] P c =G c c pa (T2-T1)

[0111] Where: G c is the compressor inlet air flow, c pa is the specific heat of air at constant pressure.

[0112] The combustion chamber module characterizes the dynamic response characteristics of the outlet temperature and the variation law of the compression loss through the principle of conservation of mass and energy. During operation, the high-pressure air output by the compressor is mixed with the fuel and burned in the combustion chamber to form high-temperature combustion gas to drive the turbine. It is worth noting that the composite cavity composed of the compressor exhaust diffuser and the combustion chamber has a significant volume effect, forming a volume module: when the working conditions are switched, the transient fluctuations of the thermodynamic parameters of the gas in the cavity will lead to changes in the storage mass, thereby causing non-steady-state differences in the working fluid flow rate at the inlet and outlet of the system. In order to establish a high-precision dynamic model, it is necessary to focus on the mass inertia effect formed by the cavity, but the thermal inertia effect caused by the change in the temperature field can be ignored.

[0113] In a single-shaft gas turbine (compressor, combustor, and turbine are coaxially connected), the volumetric effects described above are primarily reflected in the combustor module. The gas turbine's volume dynamics module is a mathematical model used to simulate the volumetric effects of gases in the system. In gas turbine simulation, the mathematical model for volumetric effects can be derived from the mass conservation equation to form the following first-order differential equation:

[0114]

[0115] Where: V is the size of the volume module, R is the gas constant, m is the mass of the gas in the volume module, p out and T out are the gas pressure and temperature at the outlet of the volume module, G in and G out The gas flow rates at the inlet and outlet of the volume module are measured respectively.

[0116] The key to modeling the combustion chamber module is the trend of the combustion chamber outlet temperature and the change of the combustion chamber pressure loss. Combining the principles of conservation of mass and energy, for the ideal gas state equation ρ=p / R g T, and take the derivative of both sides with respect to time t. Combining the above first-order differential equations, the differential equations for the dynamic process of the combustion chamber module are expressed as follows:

[0117]

[0118] Where: G2, G f G3 and G4 are respectively the air flow rate at the combustion chamber inlet, the fuel flow rate, and the gas flow rate at the combustion chamber outlet; p3 and T3 are respectively the pressure and temperature at the combustion chamber outlet; V is the combustion chamber volume; R g is the fuel gas constant, η B is the combustion efficiency of the combustion chamber; h2, H u , h3 are the air enthalpy at the combustion chamber inlet, the fuel lower calorific value, and the gas enthalpy at the combustion chamber outlet, respectively. The dynamic changes in the combustion efficiency of the combustion chamber under different loads directly affect the transient response of the system (such as temperature, pressure, fuel consumption, etc.). By quantifying the efficiency η B The relationship between the model and the load can avoid the error caused by the steady-state assumption and make the transient model closer to the actual working conditions.

[0119] This paper uses a physical model fusion method based on the combustion dynamics equation and thermodynamic model to introduce a load-related correction coefficient. Introducing the air load parameter as a correction factor, the following combustion chamber combustion efficiency formula is obtained:

[0120] η B =-5.47×10 -11 L 5 +3.98×10 -8 L 4 -8.74×10 -6 L 3 +3.00×10 -4 L 2 -4.57×10 -3 L+99.7

[0121] Where: L is the air load parameter of the combustion chamber, which is defined as follows:

[0122]

[0123] Similar to the compressor module, the turbine module is based on the turbine characteristic curve and uses the turbine expansion ratio π T and turbine efficiency η T Correlation calculation of turbine output power P T, and integrates the effect of cooling air flow on work, reflecting the performance under different working conditions. The turbine characteristic curve provides the turbine equivalent flow About turbine expansion ratio π T =p3 / p4, turbine equivalent speed The functional relationship f3, and the turbine efficiency η T About turbine expansion ratio π T =p3 / p4, turbine equivalent speed Function relationship f4:

[0124]

[0125] The outlet temperature of the turbine module is:

[0126]

[0127] Where: T3 and T4 are the turbine inlet temperature and turbine outlet temperature respectively; π T and η T are turbine expansion ratio and turbine efficiency respectively; k g is the gas specific heat ratio. After determining the turbine outlet temperature, the turbine expansion work power is:

[0128] P T =G T c pg (T3-T4)

[0129] Where: G T is the gas flow at turbine inlet, c pg is the specific heat capacity of the gas at constant pressure.

[0130] The compressor and turbine modules are developed based on the operating condition of constant rotor speed n. There are three factors that determine the operation of the gas turbine rotor: the torque M generated by the turbine T , the counter torque M generated by the compressor and generator C and M G , friction torque M f The unbalanced torque above causes the gas turbine rotor to generate angular acceleration (deceleration). Therefore, the rotor dynamics module is used to simulate and analyze the dynamic behavior of the gas turbine rotor.

[0131] The rotor dynamics equation for the gas turbine rotor moment of inertia is:

[0132]

[0133] Since ω=πn / 30 and P=Mω, the rotor dynamics equation can be transformed into:

[0134]

[0135] Where: ω is the rotor angular velocity, n is the gas turbine speed, I is the rotor moment of inertia, P T 、P c 、P G 、P f They are respectively the turbine work power, compressor work power, generator power and the power generated by friction torque.

[0136] Step 3: Determine the key influencing parameters that are significantly sensitive to thermodynamic efficiency and fluid dynamic characteristics, and determine the correction factors based on the key influencing parameters to optimize the model accuracy.

[0137] Before calibrating the gas turbine transient model, a global sensitivity analysis is performed to identify the key parameters that control gas turbine performance, which are then prioritized for optimization as correction factors to improve correction efficiency. Global sensitivity analysis is suitable for scenarios with large parameter ranges, interactions, or nonlinear responses, which are characteristics of gas turbines. By adjusting the parameter changes of the correction factors, the sensitivity differences of the parameters to the gas turbine output results are analyzed. Step 3 includes the following steps 301 to 303:

[0138] Step 301 : determining key influencing parameters that are significantly sensitive to thermodynamic efficiency and fluid dynamics characteristics.

[0139] Priority is given to key influencing parameters that are significantly sensitive to thermodynamic efficiency (such as combustion and energy conversion) and fluid dynamic characteristics (such as pressure distribution). These key influencing parameters include compressor efficiency, combustor combustion efficiency, turbine efficiency, compressor inlet air flow, combustor total pressure recovery coefficient, and mechanical efficiency.

[0140] Step 302 : Apply a dimensionless product term to the determined key influencing parameters, assign a proportionality coefficient to the key influencing parameters, and obtain a potential deviation range of the key influencing parameters under actual working conditions.

[0141] The interval of the proportional coefficient is [0.5, 1.0].

[0142] Step 303 : sort the key influencing parameters according to the influence quantification and sensitivity, and use the key influencing parameters whose potential deviation range meets the deviation threshold as correction factors of the gas turbine transient model.

[0143] Using parameter perturbation and global sensitivity analysis, we quantify the impact of these key influencing parameters on gas turbine output parameters (such as gas turbine output power and turbine outlet temperature) when they vary within the range of [0.5, 1.0] multiplied by a dimensionless scaling factor. Key influencing parameters with significant impact on output parameters are selected as correction factors for the gas turbine transient model.

[0144] Step 4, calculating the correction factor of the gas turbine transient model determined in step 3. Step 4 includes the following steps 401 to 402:

[0145] Step 401: Establish the following calculation formula for the correction factor of the gas turbine transient model:

[0146] findx i i=1,2,3,...,m

[0147]

[0148] st 0.5≤x i ≤1.1

[0149] Among them, x i is the gas turbine transient model correction factor to be calculated (the key influencing parameter obtained from the sensitivity analysis in the previous step), is the simulation result calculated by the transient mechanism model, is the actual operating data measured by the gas turbine, i and j represent the input parameters and the number of corresponding variables in the optimization problem, respectively. find indicates finding the optimal correction factor, min indicates minimization, obj indicates the objective function, and st indicates the constraints.

[0150] Step 402 : Calculate the correction factor of the gas turbine transient model using a gradient descent algorithm, a genetic algorithm, an ant colony algorithm, or a cuckoo search algorithm.

[0151] In the process of optimizing the correction factor of the transient model of gas turbine, a variety of local and global optimization methods are introduced to replace the traditional single algorithm, including four optimization methods: Gradient Descent Algorithm (GDA), Genetic Algorithm (GA), Ant Colony Optimization (ACO) and Cuckoo Search Algorithm (CSA).

[0152] GDA is an iterative optimization algorithm for minimizing the objective function. Its core idea is to gradually adjust the parameters in the opposite direction of the objective function gradient until the minimum value (or close to the minimum value) of the function is found.

[0153] GA is a heuristic optimization algorithm based on biological evolution mechanism (natural selection, genetic variation), which searches for the optimal solution by simulating the evolutionary process of "survival of the fittest".

[0154] ACO is a swarm intelligence optimization algorithm inspired by the foraging behavior of ants in nature. Real ants release pheromones to mark paths, and other ants tend to choose paths with high pheromone concentrations, forming a positive feedback mechanism that ultimately leads to the shortest path from the nest to food.

[0155] CSA is a swarm intelligence optimization algorithm inspired by the parasitic reproduction behavior of cuckoos and Levy's flight (random walk). Figure 3 A flowchart comparing the core steps of the four optimization algorithms in the context of model correction is given. Step 402 includes the following steps 4021 to 4024:

[0156] Step 4021 , initializing key algorithm parameters in the gradient descent algorithm, genetic algorithm, ant colony algorithm or cuckoo search algorithm.

[0157] Key algorithm parameters of the gradient descent algorithm, including learning rate and batch size;

[0158] The key parameters of the genetic algorithm include population size, crossover probability, and mutation probability;

[0159] The key parameters of the ant colony algorithm include pheromone weight, heuristic information weight, pheromone volatility rate, and ant colony size;

[0160] The key algorithm parameters of the cuckoo search algorithm include discovery probability, step factor (LevyFlight parameter), and population size.

[0161] Step 4022: perform fitness evaluation on the value of the objective function obj.

[0162] The fitness function is the core component of the optimization algorithm, which is used to quantify the quality of candidate solutions and directly guide the algorithm's search direction.

[0163] The gradient descent algorithm GDA and the cuckoo search algorithm CSA directly use the objective function value as the fitness function.

[0164] Genetic Algorithm GA usage as a fitness function.

[0165] Ant Colony Algorithm ACO Selection As the fitness function, to select the path, where τ is the pheromone concentration of path i→j, η ij is the heuristic factor, and α and β are the adjustment weight parameters.

[0166] Step 4023: Update the parameters of the gradient descent algorithm, genetic algorithm, ant colony algorithm, or cuckoo search algorithm. The specific parameters of each method are as follows.

[0167] GDA: By calculating the gradient of the loss function with respect to the parameters, and gradually adjusting the parameters in the opposite direction of the gradient to minimize the loss function.

[0168] GA: Crossover: Swapping some genes of two parent individuals to generate new individuals. Mutation: Randomly changing individual genes with low probability to maintain population diversity.

[0169] ACO: Pheromone Evaporation: All pheromones on all paths decay at a certain rate to avoid local optima. Pheromone Reinforcement: High-quality paths (such as the shortest path) are reinforced by ants.

[0170] CAS: Levy Flight: A new generation of solutions is generated through Levy Flight random walks, whose step size follows a heavy-tailed distribution, balancing local refinement with global jumps. Best-in-Class: Retain high-quality solutions and eliminate those with poor fitness.

[0171] Step 4024: terminate when the termination condition is met, and obtain the value of the correction factor of the gas turbine transient model.

[0172] The termination condition is that the maximum number of iterations or the fitness threshold reaches convergence stability.

[0173] Step 5: Use the calculated correction factor to adjust the gas turbine transient model to obtain a corrected gas turbine transient model.

[0174] Using the calculated correction factor, write the corrected factor into the input file of the gas turbine transient model, multiply it by the original value, run the transient simulation model, record the key outputs (such as gas turbine output power P, turbine outlet temperature T4), and compare the simulation output with the experimental data results. If the error still exceeds the threshold after correction, it is necessary to return to the optimization algorithm to adjust the correction factor range or increase the sensitive parameters.

[0175] The following is an example of a heavy-duty gas turbine in a power plant.

[0176] This case uses sensor data obtained from a heavy-duty gas turbine operating at a power plant on November 1, 2024. This dataset includes key parameters affecting the gas turbine's aerodynamic performance, including temperature, pressure, speed (n), and fuel flow rate. The sampling interval is 1 minute and covers the entire operating cycle from cold start to steady-state operation. Figure 4 It shows that the input variables of the gas turbine transient process include ambient pressure, ambient temperature, fuel flow rate, IGV angle change and other parameters. Figure 5The time evolution of output power and speed is shown, including three key operating benchmarks: startup process, part load process and steady-state process.

[0177] Based on professional knowledge and historical experience, the parameters that have a significant impact on thermodynamic efficiency and fluid dynamic characteristics are adjusted first, including compressor efficiency, combustor combustion efficiency, turbine efficiency, compressor inlet air flow, combustor total pressure recovery coefficient, and mechanical efficiency. Each key influencing parameter is multiplied by a dimensionless parameter of [0.5, 1.0] to quantify the operational impact of each key influencing parameter change on the gas turbine output results. The relationship between the dual-axis diagram (gas turbine output power (Power) and turbine outlet temperature (TOT)) and the sensitivity output of the key influencing parameters is shown as follows: Figure 6 The quantitative sensitivity indexes are listed in Table 1. First, the improvement of gas turbine power output (Power) is related to the compressor efficiency (1.19×10 5 ), combustion chamber combustion efficiency (0.95×10 5 ) and turbine efficiency (1.87×10 5 ) is closely related to the improvement of turbine outlet temperature (TOT). Secondly, the trend of turbine outlet temperature (TOT) is closely related to the compressor efficiency (↓1.89×10 2 ), turbine (↓2.95×10 2 ), compressor inlet air flow (↓4.08×10 2 ) and the combustion chamber total pressure recovery coefficient (↓1.28×10 2 ) and is inversely proportional to the combustion efficiency of the combustion chamber (↑2.16×10 2 ) maintains a positive correlation. Therefore, the correction factors are selected as follows: compressor efficiency, combustor combustion efficiency, turbine efficiency, compressor inlet air flow rate, and combustor total pressure recovery coefficient.

[0178] Table 1 Effect of key parameters on gas turbine power output (Power) and turbine outlet temperature (TOT)

[0179]

[0180] Set up and initialize four optimization algorithms, and the parameter configuration is shown in Table 2. c and P m are the probabilities of recombination and perturbation, ρ is the evaporation coefficient, α and β are the parameters controlling the influence of pheromone and heuristic, respectively, P a is the probability of discovery.

[0181] Table 2 Parameter configuration of four optimization algorithms

[0182]

[0183] exist Figure 5Points A, B, and C were selected as representative points, representing the startup process, the partial load transient process, and the base load steady-state process, respectively. The optimization algorithm described above was used to solve for the optimization factor. As the number of iterations increased, the optimization algorithm continuously adjusted the model parameters, gradually decreasing the loss value and converging, ultimately approaching the minimum value of the objective function. The error trends of the four optimization algorithms for the startup, partial load, and base load optimization iterations all decreased with increasing iterations, demonstrating the excellent performance of the optimization algorithms. Figure 7 The optimization process of the genetic algorithm at the base load point is presented. Table 3 shows the loss values of the four optimization algorithms after 100 optimization iterations at the three points. Table 3 shows that the genetic algorithm consistently achieves the smallest error during the optimization process, outperforming the other three algorithms.

[0184] Table 3 Loss results of the four methods after 100 optimization iterations at three points

[0185] Operational phase GDA GA ACO CSA start up 535.15 94.79 799.38 363.90 Partial load 11138.80 92.59 2152.36 75.07 Base load 16479.60 73.53 1945.22 1066.10

[0186] Table 4 Comparison of the results of the three operation stages using four optimization methods

[0187]

[0188]

[0189] Table 4 summarizes the quantitative comparison results of the four methods for the four key indicators. For each operation stage, the actual data, the original model simulation results and the simulation results corrected by the four optimization algorithms and their relative errors are shown in Table 4. (where y r is the simulation result, y a is the actual data). From Table 3, we can observe that the model corrected by GA has an average reduction of 86.3% in the three running stages (calculation formula: Among them, RE O is the relative error between the original uncorrected simulation data and the actual data, RE GA is the relative error between the simulation data and the actual data after correction using the GA optimization method), especially the compressor outlet pressure at startup (RE: -12.57% → -1.14×10 -3 %), as well as the accuracy of part-load output power (RE:-24.99%→0.07%) and base-load output power (RE:22.82%→0.12%) are better than those of other methods.

[0190] Therefore, during the entire operation phase Figure 4 and Figure 5 As shown, a genetic algorithm is used to improve the correction factor of the mechanism transient model. Figure 8 The transient performance model modified by GA and the actual measured data are given at the compressor outlet pressure ( Figure 8 a) Turbine outlet pressure ( Figure 8 b) Compressor outlet temperature ( Figure 8 c) Turbine outlet temperature ( Figure 8 d) Gas turbine output power ( Figure 8 e). The solid line and dashed line represent the actual measurement data and model simulation results, respectively. Figure 8 a-8b shows the good consistency between the pressure trajectory obtained by the model and the actual pressure trajectory during the entire startup process, accurately capturing the transient pressure dynamic changes. Figure 8 c-8d show the consistency of temperature transient changes. Figure 8 f shows the relative errors of the above parameters (less than 10%). These results verify the accuracy of the GA correction model in the transient operation simulation of gas turbines.

[0191] This example compares data from two actual gas turbines during different operating phases (including startup, load ramping, and steady-state transition). Quantitative verification results show that the average relative error of the model using the correction method is reduced by 86.3% compared to the original transient model, significantly improving model accuracy. The relative error between the corrected parameter results and the actual data is less than 1% (steady state) and 10% (transient). This simultaneous improvement in model accuracy, real-time performance, and robustness supports the efficient design and control of complex systems.

[0192] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications to the technical solutions described in the above embodiments, or equivalent replacement of some or all of the technical features therein, do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automatic correction of gas turbine transient model based on mathematical and physical fusion, characterized by: The following processes are included: Step 1: preprocessing the gas turbine operating data; Step 2, establishing a gas turbine transient model; The gas turbine transient model includes: a compressor module, a combustion chamber module, a turbine module and a rotor dynamics module; The performance of the compressor module is represented by a compressor characteristic curve diagram; the compressor characteristic curve diagram provides the compressor equivalent flow About compressor pressure ratio π c =p2 / p1, compressor reduced speed The functional relationship between the inlet guide vane angle f1 and the compressor efficiency η c About pressure ratio π c =p2 / p1, converted speed And the functional relationship f2 of the inlet guide vane angle IGV: The compressor reduced flow rate and compressor efficiency can be obtained through the functional relationship between the compressor pressure ratio, the compressor reduced speed and the IGV. By determining only three of the five parameters, the working state of the compressor can be determined. The outlet temperature of the compressor module is: Where T1 and T2 are the compressor inlet temperature and compressor outlet temperature respectively; π c and η c are compressor pressure ratio and compressor efficiency respectively; k a is the air specific heat ratio; After determining the compressor outlet temperature, the power consumed by the compressor module is: P c =G c ·c pa ·(T2-T1) Where G c is the compressor inlet air flow, c pa is the specific heat capacity of air at constant pressure; The combustion chamber module characterizes the dynamic response characteristics of the outlet temperature and the variation law of the compression loss through the principles of conservation of mass and energy. In the gas turbine simulation, the mathematical model of the volume effect can be obtained by the mass conservation equation to obtain the following first-order differential equation: Where V is the size of the volume module, R is the gas constant, m is the mass of the gas in the volume module, and p is the volume of the gas. out and T out are the gas pressure and temperature at the outlet of the volume module, G in and G out The gas flow rates at the inlet and outlet of the volume module respectively; The differential equations of the combustion chamber module dynamic process are expressed as follows: In the formula, G2, G f G3 and G4 are respectively the air flow rate at the combustion chamber inlet, the fuel flow rate, and the gas flow rate at the combustion chamber outlet; p3 and T3 are respectively the pressure and temperature at the combustion chamber outlet; V is the combustion chamber volume; R g is the fuel gas constant, η B is the combustion efficiency of the combustion chamber; h2, H u and h3 are the enthalpy of air at the combustion chamber inlet, the lower calorific value of fuel, and the enthalpy of gas at the combustion chamber outlet, respectively; Introducing the air load parameter as a correction factor, the following combustion chamber combustion efficiency formula is obtained: η B =-5.47×10 -11 L 5 +3.98×10 -8 L 4 -8.74×10 -6 L 3 +3.00×10 -4 L 2 -4.57×10 -3 L+99.7 Where: L is the air load parameter of the combustion chamber, which is defined as follows: The turbine module is based on the turbine characteristic curve, through the turbine expansion ratio π T and turbine efficiency η T Correlation calculation of turbine output power P T , and integrates the effect of cooling air flow on work, reflecting the performance under different working conditions; the turbine characteristic curve provides the turbine equivalent flow About turbine expansion ratio π T =p3 / p4, turbine equivalent speed The functional relationship f3, and the turbine efficiency η T About turbine expansion ratio π T =p3 / p4, turbine equivalent speed Function relationship f4: The outlet temperature of the turbine module is: Where T3 and T4 are the turbine inlet temperature and turbine outlet temperature respectively; π T and η T are turbine expansion ratio and turbine efficiency respectively; k g is the gas specific heat ratio; after determining the turbine outlet temperature, the turbine expansion work power is: P T =G T ·c pg ·(T3-T4) Where G T is the gas flow at turbine inlet, c pg is the specific heat capacity of gas at constant pressure; The rotor dynamics module is used to simulate and analyze the dynamic behavior of the gas turbine rotor. The rotor dynamics equation for the gas turbine rotor moment of inertia is: Since ω=πn / 30 and P=Mω, the rotor dynamics equation can be transformed into: Where ω is the rotor angular velocity, n is the gas turbine speed, I is the rotor moment of inertia, and P is T 、P c 、P G 、P f They are respectively the turbine work power, compressor work power, generator power and power generated by friction torque; Step 3: determine the key influencing parameters that are significantly sensitive to the thermodynamic efficiency and fluid dynamic characteristics, and determine the correction factors based on the key influencing parameters; Step 4, calculating the correction factor of the gas turbine transient model determined in step 3; Step 5: Use the calculated correction factor to adjust the gas turbine transient model to obtain a corrected gas turbine transient model.

2. The automatic correction method for gas turbine transient model based on mathematical-physical fusion according to claim 1 is characterized in that: The operating data of the gas turbine, including thermal parameters, mechanical parameters, performance index data, control system data, and environmental and weather data; The thermal parameters include compressor inlet temperature T1, compressor outlet temperature T2, compressor inlet pressure p1, compressor outlet pressure p2, and turbine outlet temperature T4; The mechanical parameters include the rotational speed n of the rotor; The performance index data includes the power generation P G , fuel flow G f ; The control system data includes valve opening, IGV angle, and fuel distribution ratio f; The environment and weather data include ambient temperature T0, ambient pressure P0, and ambient relative humidity HR.

3. The automatic correction method for gas turbine transient model based on mathematical-physical fusion according to claim 2 is characterized in that: The data preprocessing includes data cleaning; The data cleaning includes the following processes: Use sliding window interpolation or dynamic time warping to unify timestamps and synchronize data; Perform data outlier detection and missing value interpolation; The Bayesian wavelet packet method is used to reduce the noise of the data.

4. The automatic correction method for gas turbine transient model based on mathematical-physical fusion according to claim 1 is characterized in that: Step 3 includes the following steps 301 to 303: Step 301: determining key influencing parameters that are significantly sensitive to thermodynamic efficiency and fluid dynamic characteristics; the key influencing parameters include: compressor efficiency, combustor combustion efficiency, turbine efficiency, compressor inlet air flow, combustor total pressure recovery coefficient, and mechanical efficiency; Step 302 , applying a dimensionless product term to the determined key influencing parameters, assigning a proportionality coefficient to the key influencing parameters, and obtaining a potential deviation range of the key influencing parameters under actual working conditions; Step 303 : sort the key influencing parameters, and use the key influencing parameters whose potential deviation ranges meet the deviation threshold as correction factors for the gas turbine transient model.

5. The automatic correction method for gas turbine transient model based on mathematical-physical fusion according to claim 1 is characterized in that: Step 4 includes the following steps 401 to 402: Step 401: Establish the following calculation formula for the correction factor of the gas turbine transient model: findx i i=1,2,3,...,m s.t.0.5≤x i ≤1.1 Among them, x i is the gas turbine transient model correction factor to be calculated, is the simulation result calculated by the transient mechanism model, is the actual operating data measured by the gas turbine, i and j represent the input parameters of the optimization problem and the number of corresponding variables, respectively, find means finding the best correction factor, min means minimization, obj means objective function, and st means constraint condition; Step 402 : Calculate the correction factor of the gas turbine transient model using a gradient descent algorithm, a genetic algorithm, an ant colony algorithm, or a cuckoo search algorithm.

6. The automatic correction method for gas turbine transient model based on mathematical-physical fusion according to claim 5 is characterized in that: Step 402 includes the following steps 4021 to 4024: Step 4021, initializing key algorithm parameters in the gradient descent algorithm, genetic algorithm, ant colony algorithm, or cuckoo search algorithm; Step 4022, performing fitness evaluation on the value of the objective function obj; The gradient descent algorithm and the cuckoo search algorithm directly use the objective function value as the fitness function; Genetic Algorithm Use As a fitness function; Ant colony algorithm selection As the fitness function, to select the path, where τ ij is the pheromone concentration of path i→j, η ij is the heuristic factor, α and β are the adjustment weight parameters; Step 4023, updating parameters in the gradient descent algorithm, genetic algorithm, ant colony algorithm, or cuckoo search algorithm; Step 4024: terminate when the termination condition is met, and obtain the value of the correction factor of the gas turbine transient model.

Images